I want to report progress I've made tonight on an extended form of the
OVS sandbox that I expect to be useful for testing OVN. The main point
of the extension is to make it easy to start up multiple sandboxed
instances of OVS and then connect them together.
What I have is at:
https://gith
On Mon, Jun 8, 2015 at 3:50 PM, Pravin B Shelar wrote:
> Some of code is backported from following commit.
>
> commit 13dd4a9738e99684a56b10ce2f1a5ee2d2ec2f9f
> Author: Joe Stringer
> Date: Tue Mar 24 16:16:18 2015 -0700
>
> compat: Fix RHEL7 build.
>
> Tested against 3.10.0-229.el7.x86
On Tue, Jun 09, 2015 at 06:19:35PM -0300, Thadeu Lima de Souza Cascardo wrote:
> On Tue, Jun 09, 2015 at 03:03:03PM -0300, Flavio Leitner wrote:
> > On Mon, Jun 08, 2015 at 01:05:41PM -0300, Thadeu Lima de Souza Cascardo
> > wrote:
> > > Support IGMPv3 messages with multiple records. Make sure all
We formerly converted tunnel attributes to their flow representation
before formatting but now perform all operations directly on the
netlink attributes.
There is one remaining use of odp_tun_key_from_attr() that is not
used for the purposes of generating a flow. This is to check the
mask but this
On Wed, Jun 10, 2015 at 03:13:21PM +, Daniele Di Proietto wrote:
>
>
> On 10/06/2015 12:48, "Gray, Mark D" wrote:
> >
> >The vhost port won't generate an RSS hash because it is a virtual NIC.
> >
> >>
> >
> >> It doesn't cause a problem, just make the pkt fall into a slow path,
> >>should w
On 2015/6/10 23:34, Gray, Mark D wrote:
On 10/06/2015 12:48, "Gray, Mark D" wrote:
The vhost port won't generate an RSS hash because it is a virtual NIC.
It doesn't cause a problem, just make the pkt fall into a slow path,
should we
fix it?
Thanks for investigating this. We should defin
Some DB modification transactions won't affect the columns of a monitor.
Skip those updates.
Signed-off-by: Andy Zhou
---
ovsdb/monitor.c | 68 -
1 file changed, 67 insertions(+), 1 deletion(-)
diff --git a/ovsdb/monitor.c b/ovsdb/monitor.
Currently, JSON cache is always flushed whenever the monitor receives
a transaction. This patch improves the JSON cache efficiency by only
flush the JSON cache when a transaction causes client visible
changes, avoiding unnecessary flushes.
Suggested-by: Ben Pfaff
Signed-off-by: Andy Zhou
---
ov
The logic of determining row update type will be useful in the following
patch. Make it into a function.
Signed-off-by: Andy Zhou
---
ovsdb/monitor.c | 20
1 file changed, 16 insertions(+), 4 deletions(-)
diff --git a/ovsdb/monitor.c b/ovsdb/monitor.c
index 257959c..107bdef
> On Jun 10, 2015, at 5:43 PM, Ben Pfaff wrote:
>
> On Tue, Jun 09, 2015 at 05:24:17PM -0700, Jarno Rajahalme wrote:
>> The final flow stats are available only after there are no references
>> to the rule. Postpone sending the flow removed message until the
>> final stats are available.
>>
>>
On Wed, Jun 10, 2015 at 05:52:21PM -0700, Jarno Rajahalme wrote:
>
> > On Jun 10, 2015, at 5:46 PM, Ben Pfaff wrote:
> >
> > On Tue, Jun 09, 2015 at 05:24:18PM -0700, Jarno Rajahalme wrote:
> >> struct match in ofputil_flow_mod uses a lot of space, when used for a
> >> stored bundle message. Th
> On Jun 10, 2015, at 5:46 PM, Ben Pfaff wrote:
>
> On Tue, Jun 09, 2015 at 05:24:18PM -0700, Jarno Rajahalme wrote:
>> struct match in ofputil_flow_mod uses a lot of space, when used for a
>> stored bundle message. This patch adds a new struct
>> ofputil_miniflow_mod, that uses a minimatch ins
On Tue, Jun 09, 2015 at 05:24:19PM -0700, Jarno Rajahalme wrote:
> Add support for port mods in an OpenFlow 1.4 bundle, as required for
> the minimum support level by the OpenFlow 1.4 specification. If the
> bundle includes port mods, it may not specify the OFPBF_ATOMIC flag.
> Port mods and flow
On Tue, Jun 09, 2015 at 05:24:18PM -0700, Jarno Rajahalme wrote:
> struct match in ofputil_flow_mod uses a lot of space, when used for a
> stored bundle message. This patch adds a new struct
> ofputil_miniflow_mod, that uses a minimatch instead of match in hopes
> of using less memory when handlin
On Tue, Jun 09, 2015 at 05:24:17PM -0700, Jarno Rajahalme wrote:
> The final flow stats are available only after there are no references
> to the rule. Postpone sending the flow removed message until the
> final stats are available.
>
> Signed-off-by: Jarno Rajahalme
I'm a little nervous about
On Tue, Jun 09, 2015 at 05:24:16PM -0700, Jarno Rajahalme wrote:
> Handling evictions was broken in the previous patches. Eviction took
> place early in the commit, and actually inappropriately bumped the
> version number too early. Now eviction is treated much like a flow
> modification, where a
On Tue, Jun 09, 2015 at 05:24:15PM -0700, Jarno Rajahalme wrote:
> Classifier's rule count now contains temporary duplicates and rules
> whose deletion has been deferred. Maintain a new 'n_flows' count in
> struct oftable to as the count of rules in the latest version.
>
> Signed-off-by: Jarno Ra
On Tue, Jun 09, 2015 at 05:24:14PM -0700, Jarno Rajahalme wrote:
> Each rule is now added or deleted in a specific tables version. Flow
> tables are versioned with a monotonically increasing 64-bit integer,
> where positive values are valid version numbers.
>
> Rule modifications are implemented
On Tue, Jun 09, 2015 at 05:24:13PM -0700, Jarno Rajahalme wrote:
> Signed-off-by: Jarno Rajahalme
In two places, the comment
/* Set tables version after the bridge has been fixed. */
took me a while to follow ("Why was the bridge broken?"). Maybe:
/* The bridge is now known so
On Tue, Jun 09, 2015 at 05:24:12PM -0700, Jarno Rajahalme wrote:
> Signed-off-by: Jarno Rajahalme
Tests, tests, tests, I love tests.
Tests, tests, tests, I love tests.
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org
On Tue, Jun 09, 2015 at 05:24:11PM -0700, Jarno Rajahalme wrote:
> The traversal of the list of identical rules from the lookup threads
> is fragile in the list head is removed during the list traversal.
fragile "if" the list head...?
> This patch simplifies the implementation of that list by mak
> On Jun 10, 2015, at 3:51 PM, Ben Pfaff wrote:
>
> On Tue, Jun 09, 2015 at 05:24:10PM -0700, Jarno Rajahalme wrote:
>> This patch allows classifier rules to become visible and invisible in
>> specific versions. A 'version' is defined as a positive monotonically
>> increasing integer, which nev
> On Jun 10, 2015, at 3:57 PM, Ben Pfaff wrote:
>
> On Wed, Jun 10, 2015 at 03:51:05PM -0700, Ben Pfaff wrote:
>> On Tue, Jun 09, 2015 at 05:24:10PM -0700, Jarno Rajahalme wrote:
>>> This patch allows classifier rules to become visible and invisible in
>>> specific versions. A 'version' is defi
On Wed, Jun 10, 2015 at 03:51:05PM -0700, Ben Pfaff wrote:
> On Tue, Jun 09, 2015 at 05:24:10PM -0700, Jarno Rajahalme wrote:
> > This patch allows classifier rules to become visible and invisible in
> > specific versions. A 'version' is defined as a positive monotonically
> > increasing integer,
On Tue, Jun 09, 2015 at 05:24:10PM -0700, Jarno Rajahalme wrote:
> This patch allows classifier rules to become visible and invisible in
> specific versions. A 'version' is defined as a positive monotonically
> increasing integer, which never wraps around.
>
> The new 'visibility' attribute repla
Ben,
I decided to drop this patch as I wrote a bug fix relating to this. So this
patch is superseded by the later patch.
I did not investigate the potential code generation differences, but decided
that we need not worry about other threads, when the documentation says that
the function may on
Postponed 'next' member poisoning was based on the faulty assumption
that postponed functions would be called in the order they were
postponed. This assumption holds only for the functions postponed by
any single thread. When functions are postponed by different
threads, there are no guarantees o
> On Jun 10, 2015, at 9:22 AM, Ben Pfaff wrote:
>
> On Tue, Jun 09, 2015 at 05:24:08PM -0700, Jarno Rajahalme wrote:
>> Weirdest things can bother you at night when you try to sleep ;-)
>>
>> Now we have function triples such as add_flow_begin(),
>> add_flow_finish(), and add_flow_revert(), whe
> On Jun 10, 2015, at 11:29 AM, Ben Pfaff wrote:
>
> On Tue, Jun 09, 2015 at 05:24:09PM -0700, Jarno Rajahalme wrote:
>> All existing ovs-ofctl flow mod commands now take an optional
>> '--bundle' argument, which executes the flow mods as a single
>> transaction. OpenFlow 1.4+ is implicitly ass
I've been doing some thinking about OpenStack Neutron's provider
networks and how we might be able to support that with OVN as the
backend for Neutron. Here is a start. I'd love to hear what others think.
Provider Networks
=
OpenStack Neutron currently has a feature referred to
On Tue, Jun 09, 2015 at 05:24:09PM -0700, Jarno Rajahalme wrote:
> All existing ovs-ofctl flow mod commands now take an optional
> '--bundle' argument, which executes the flow mods as a single
> transaction. OpenFlow 1.4+ is implicitly assumed when '--bundle' is
> specified.
>
> ovs-ofctl 'add-fl
On Wed, Jun 10, 2015 at 07:23:36PM +0300, Andrey Korolyov wrote:
> > Thanks for the review. I applied these patches to master.
> >
> > I don't really understand some of your comments here. If you could
> > express them in the form of a patch, then it would make it clearer and
> > I'd be able to a
On Tue, Jun 09, 2015 at 05:24:08PM -0700, Jarno Rajahalme wrote:
> Weirdest things can bother you at night when you try to sleep ;-)
>
> Now we have function triples such as add_flow_begin(),
> add_flow_finish(), and add_flow_revert(), where a modification is
> started in *_begin(), which can fail
>
> Thanks for the review. I applied these patches to master.
>
> I don't really understand some of your comments here. If you could
> express them in the form of a patch, then it would make it clearer and
> I'd be able to apply it directly.
>
Thanks Ben, hopefully I`ll propose related fix in dp
On Mon, Jun 08, 2015 at 12:34:25AM +0300, Andrey Korolyov wrote:
> On Sun, Jun 7, 2015 at 11:14 PM, Ben Pfaff wrote:
> > On Sun, Jun 07, 2015 at 10:15:29PM +0300, Andrey Korolyov wrote:
> >> On Sun, Jun 7, 2015 at 7:48 PM, Ben Pfaff wrote:
> >> > Setting CFLAGS by hand before invoking dpkg-buildf
A long time ago, the Open vSwitch build did not depend on Python (whereas
the runtime did), so the "make dist" based distribution included the
results of Python build tools. Later, the build began building on Python,
but the distribution still included some of those results, because no one
had gon
v1->v2: Correct statements about XenServer/XCP, by stating that versions
with Python 2.4 work if you install Python 2.7 on them.
Ben Pfaff (2):
Increase prerequisite from Python 2.4 to Python 2.7.
Makefiles: Stop distributing files because building them requires
Python.
COPYING
OK, that sounds like a step forward.
I guess that means that we can start using Python 2.7 immediately,
then. I'll respin my patches.
On Tue, Jun 09, 2015 at 08:06:52AM -0400, Andy Hill wrote:
> Hi Ben,
>
> XenServer 6.6 (still to be released) will ship with Python 2.7 as the
> system python. N
>
>
>
> On 10/06/2015 12:48, "Gray, Mark D" wrote:
> >
> >The vhost port won't generate an RSS hash because it is a virtual NIC.
> >
> >>
> >
> >> It doesn't cause a problem, just make the pkt fall into a slow path,
> >>should we
> >
> >> fix it?
>
> Thanks for investigating this. We should d
On 10/06/2015 12:48, "Gray, Mark D" wrote:
>
>The vhost port won't generate an RSS hash because it is a virtual NIC.
>
>>
>
>> It doesn't cause a problem, just make the pkt fall into a slow path,
>>should we
>
>> fix it?
Thanks for investigating this. We should definitely fix it.
>
>> The fl
We've had this question a couple of times so we might as well document it.
Requested-by: Saurabh Shrivastava (सौरभ श्रीवास्तव)
Signed-off-by: Ben Pfaff
---
AUTHORS | 1 +
lib/ovsdb-idl.h | 6 +-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/AUTHORS b/AUTHORS
index 5a
>
> Hi:
> In "dp_packet_get_rss_hash", mbuf.hash.rss is returned directly. But when
> it's DPDK vhost pkt, rss isn't initialized, "dpif_netdev_packet_get_rss_hash"
> gets a indeterminate value. There is not the same issue for ovs linux vport.
> I confirmed it with gdb.
>
> *Flow 1:* 6.0.0.1 -> 6
On 2015/6/10 19:16, Traynor, Kevin wrote:
-Original Message-
From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Dongjun
Sent: Tuesday, June 9, 2015 4:36 AM
To: dev@openvswitch.org
Subject: [ovs-dev] Is the tx spinlock in __netdev_dpdk_vhost_send necessary?
This is the source c
> -Original Message-
> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Dongjun
> Sent: Tuesday, June 9, 2015 4:36 AM
> To: dev@openvswitch.org
> Subject: [ovs-dev] Is the tx spinlock in __netdev_dpdk_vhost_send necessary?
>
> This is the source code of "__netdev_dpdk_vhost_sen
Per this patch ovs-vswitchd searches for shared objects, with
a certain set of exposed routines, in a specific configurable path.
Once found, all compatible plugins are loaded and their routines are
called at appropriate points of the daemon execution.
Currently, the main purpose is to allow in
Mesmerising night is ready to be organized by splendid babes
http://cc4.co/JKLR
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Hi:
In "dp_packet_get_rss_hash", mbuf.hash.rss is returned directly. But
when it's DPDK vhost pkt, rss isn't initialized,
"dpif_netdev_packet_get_rss_hash" gets a indeterminate value. There is
not the same issue for ovs linux vport.
I confirmed it with gdb.
*Flow 1:* 6.0.0.1 -> 6.0.0.2
Break
47 matches
Mail list logo
